Let's Make Robots!

The Karpetbeest, a printable Jansen walker

walk gracefully under wind or other power
a set of parts, print finished succesfully1.09 MB
fresh out of the oven1.1 MB
closeup of the support structure skeinforge makes1.12 MB
locally grown produce1.49 MB

This is a mm scale model of Theo Jansen's Strandbeests suitable for printing with RepRap 3D printers - a work in progress.

Update April 2012: Finally, I finished a six-pairs-of-legs blue specimen for the show at ŠKUC gallery. It walks if you push it around the carpet! A 2m PVC electrician tube is provided to make this more comfortable. :) Rotating joins between printed parts are made with 3mm plastic filament, melted a little into a ball on each side. That works so well I only wish it could be used for everything!

Lots of hexnuts on M3 threaded rod keep the printed backbone parts of the bot nicely parallel and equally spaced. I tried with hotglue and thick copper wire first, but it was hard to get a good bond inside the hole and many of them came unstuck after a while.

Crankshaft pins are held in little holes with superglue. They like to break off and are now the weakest link. I will try epoxy next, and if that doesn't help, switch from straight pins to U or clamp shaped ones which will not "pull out of the hole". Some redesign of the printed part will be necessary for that. Or maybe the plastic filament would work here too, and I'm just not brave enough to try?

The model is built in Blender with lots of boolean modifiers using python scripting. Blender lets me see the motion and assembly before printing and because it's a script it's easy to adjust some parameters after test prints, for instance hole sizes which can be tricky to get right. The script takes Jansen's magic numbers as millimeter measurements and builds all the parts of the leg and frame in the right shapes and sizes.

I'm now (2012) testing new Blender versions with improved booleans code, which is much faster and more stable. But it still chokes sometimes on what I'm asking it to do. When finished, the generated model will be cleaner and more precise. Perhaps OpenSCAD would really be better for all this CSG...

The frame accepts skate bearings into which the crankshaft parts are glued, hopefully giving a smooth and precise circular motion and minimal friction. Hinge axes are 3mm metal pins, like the sled rails in optical drives (CD, DVD), which i've got a few lying around.

Expect more to come soon, as I progress toward a walking unpowered beest, then figure out the propulsion, and finally build a two-motor robot with 2 groups of 3 pairs == 12 legs! I'll put the design files and script up here and on Thingiverse under a free license as soon as I make sure it all fits together well, eliminate the weak spots and adjust the tolerances for good fit where needed.

a set of parts, print finished succesfully



Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Hey, did you ever get this posted up to Thingiverse?

i see it's been three months since i posted this - but i made no progress with it (had no time to make robots... shame on me) so no updated version or post on thingiverse yet. but the demo linkage was in an exhibition for a month :)) vacations are coming though so hopefully won't be too long now.

i need to beef up the "link bars" ears a bit, they break too easily as they are now. and i'm still not sure how to best fix the pins either... so really looking forward to finishing this first one. if anyone wants to collaborate on design, i can send the current files in private too.

btw, i saw theo jansen himself started selling 3d printed mini-beests on shapeways in the meantime. they look very much like the gakken kit, but require no assembly! fancy...



Great project my friend,I love 3D printers and CNC. Keep up the progress :)

I'm working on a design for printable 1-part ball bearings. Those are 608's, right? I'll beam you the STL in the near future...

These are 608 skate bearing yes, because they're easy to get.

The printed bearing should be bigger I imagine, big enough to have two shaft holes on the inner rotating piece 15mm from the center, 120 degrees apart. The outer part will be static and part of the frame, like the hole now. This way we could get rid of the two crankshaft parts, quite a bit of precision gluing and have a stronger crankshaft too, because the shaft pin could go much deeper in the hole. Now they're only a mm deep and will probably need some very strong glue.

Are your bearings completely printed, including the balls, or are you using some premade balls? I try to imagine the sound a beest with eight of those would make when walking... Hehe, remains to be heard!

Wow, ball bearings... Please keep posting. Thank you for the tags!

Yes, metal bearings, well these (608) are at least easy to get in any skate shop, which are much more common in urban centers than bearing shops these days...

Even with them, I'm not sure my legs will be able to get up to the speed you demonstrated with your wooden model, Rik!

I decided to use the bearings mainly to simplify the construction. Perhaps it would work with simple pins too, and it's also possible to print a bearing, so you'd have the bearing and crankshaft holes in a single printed piece rotating inside the race built into the frame, very elegant. But, baby steps... :)

BTW, we met and talked a bit at Campus party in Madrid, I first heard about LMR there and it grew dear to me very quickly! This really is the friendliest and hackingest robotics community on the net!

That's almost a full year ago! Who are you again? Don't remember a Slovenian Linefeed in Madrid. Ouch, the weather. And the volcano! 2010 was a very different year.


Hehe, you can catch me as the flower lover in this superslowmo video made at CP Europe, hopefully proof enough. Note the flowers were frozen with liquid nitrogen before shooting this.

I also put my face in the LMR profile, no more featureless clipart guy.

been there, done that

seen the t-shirts